Status: LOCKED. Date locked: 2026-04-24.
These five interventions have pre-committed verdicts. They exist so every other intervention I rate can be triangulated against them. If a new verdict is inconsistent with these anchors, the methodology has failed and I revisit — I do not silently bend the anchors.
Each anchor cites the external authority whose published position I am aligning with, so the calibration is auditable.
The only intervention with T1 evidence in humans on hard endpoints (all-cause mortality), replicated across thousands of cohorts and dozens of RCTs, with mechanistic plausibility across nearly every aging hallmark. Effect size: ~30% reduction in all-cause mortality at population level for moderate aerobic + resistance training vs sedentary.
Authority anchored against: WHO physical activity guidelines (2020); meta-analyses in BMJ and JAMA; Cochrane reviews of exercise for older adults.
Why this is the ceiling: if any other intervention claims "Strong," its evidence base must be at least this comprehensive. None currently are.
In mice, CR is the most replicated lifespan intervention in the history of biology — hundreds of studies across decades, multiple strains, multiple labs, dose-response established. T3 evidence with extensive replication.
Important caveat preserved in the anchor: Strong for mice. The human translation (CALERIE-2 RCT) shows biomarker improvements but no lifespan data; the rhesus monkey studies (NIA vs Wisconsin) reached different conclusions depending on control diet quality.
Authority anchored against: Richard Weindruch's body of work; NIA's CR program; CALERIE-2 trial reports.
Why this is the ceiling for mouse interventions: any new intervention claiming "Strong in mice" must match CR's level of replication, which is essentially impossible for newer interventions. So in practice, mouse-only interventions max out at "Probable."
Rapamycin has multiple ITP-positive cohorts (T3, gold standard) across both sexes (with sex-dependent effect sizes), dose-response data, and mechanistic clarity (mTOR). Human evidence is limited to T2 surrogate endpoints (PEARL trial — open-label, modest sample, biomarker outcomes; and small immunology trials showing improved vaccine response in older adults).
It is not "Strong" because human lifespan/mortality data does not exist. It is above "Suggestive" because the mouse evidence is exceptional and human surrogate data exists.
Authority anchored against: Matt Kaeberlein's published position; ITP cohort summaries; PEARL trial publications.
Why this is the anchor for "Probable": the bar to claim Probable is "T3 mouse evidence with at least suggestive human data." Rapamycin is the cleanest example. Anything I rate Probable should be defensible as having evidence at least this strong; if it doesn't, it is Suggestive.
NAD+ precursors have T4-T5 evidence: small lifespan effects in mice (single-lab, contested across labs), mechanism plausible (NAD+ decline with age is real), and human RCTs have shown NAD+ level increases without clear downstream functional benefit. Industry-funded studies dominate the literature, and several have failed to replicate independently.
Why not "Mostly hype": NAD+ biology is genuine; the mechanistic foundation is real; some non-industry data exists.
Why not "Probable": no ITP replication; mouse lifespan effects are inconsistent across labs; human RCTs lack hard or even strong surrogate endpoints.
Authority anchored against: ITP's failure to replicate NR (the closest tested precursor); Brad Stanfield's reviews; published critical reviews of NAD+ supplementation.
Why this is the anchor for "Suggestive": something interesting is probably happening, but the evidence has not climbed to the bar that Probable requires.
Resveratrol kicked off the modern longevity-supplement industry on the strength of yeast and mouse studies (T5/T4) that have largely failed to replicate at higher tiers. ITP tested it; it failed. Major mouse lifespan studies that initially showed effects had problematic controls. Human RCTs show no meaningful biomarker effect at achievable oral doses.
Authority anchored against: ITP's negative result; Aubrey de Grey's and Matt Kaeberlein's published skepticism; failed Sirtris commercialization trajectory.
Why this is the anchor for "Mostly hype": the evidence existed, it was tested rigorously, and it failed. Other interventions rated "Mostly hype" should have a similar trajectory of "popular, tested at higher tier, failed."
When rating any new intervention:
Anchors can be revised, but only via:
Revision requires a commit-message entry on this repository, re-review of every page that triangulated against the anchor, and documentation of the triggering evidence.